Development Of Multi-Agent Investment System

Development of Multi-Agent Investment System.

First of all, financial markets have become global, i.e. on the one hand, it has become very easy for investors to make investments in any world region, but on the other hand, they face a need to be sound on the specific character of each corresponding region, adapt to it and learn to work there. Secondly, financial markets have become dynamic, i.e. they can quickly change directions, thus there is a need to follow information continuously, to analyse it, to draw immediate conclusions and make investment decisions on the ground of them. Consequently,

information scattered on the internet or elsewhere, sorts out the methods of securities evaluation and finally draws conclusions. Complexity is that in order to make a decision it is necessary to coordinate the set of coherently and in parallel related processes such as information gathering, carrying out of evaluation, market surveillance etc., and to establish interrelations. Thus, providing of investment recommendation to an investor comes down to the solution of some tasks and to the control and management of separate tasks. It is assumed that it is reasonable to create such a system using software agents or rather to make multi-agent systems of them for task solution.

3. To identify and choose the suitable software agent’s platform to implement multi-agent expert investment decision support system.

a) The method of the evaluation of investor risk tolerance, the method of the portfolio property allocation and the method of the providing of investment recommendation for securities;

1. THEORETICAL AND PRACTICAL ASPECTS OF SECURITIES VALUATION, INVESTMENT DECISION MAKING AND PORTFOLIO FORMATION

number of assumptions concerning financial market which rarely come true at real

Each system is created in reliance on a certain sequence of steps which is defined by an appropriate methodology. The methodologies of multi-agent systems are divided into two groups, i.e. for general purpose and specific. Specific methodologies are general methodologies adapted for the systems completing particular, narrow tasks. Literature analysis did not provide a specific, directly applicable methodology for MEIDSS implementation. So, the author designed a new methodology combining MaSE [25], GAIA [26], JADE [27] methodologies. The substance of novelty lies in the proposal to decompose the system into integral

In the course of methodology analysis the main emphasis is laid on the identification of system objectives and preliminary model of how the system works. The objective of MEIDSS is to provide investment recommendations accentuating investor profile and act simulating the business model of investment Management Company. It makes MEIDSS unique with regard to analogical systems such as Warren, MASST and “E-investor”. MEIDSS operation principle is explained by the offered and designed horizontal and vertical system structures showed in Figure 1 and Figure 2.

Fig. 5 Cascading structure of risk tolerance evaluation expert system

Creation (design) phase required to find a new portfolio allocation decision suitable for MEIDSS. Portfolio allocation in interpreted a bit differently in